Multi-factor flame recognition method suitable for embedded platform

Abstract

The invention discloses a multi-factor flame recognition method suitable for an embedded platform. The method comprises the steps of building a fire sample library, wherein the fire sample library isfrom a network fire picture and a combustion experiment picture; obtaining a plurality of live video frames separately; extracting moving objects in each live video frame separately and obtaining oneor more quasi-fire areas; carrying out fire confirmation on one or more quasi-fire areas according to the fire sample library and judging whether fire information exists in the one or more quasi-fireareas or not; and if so, carrying out fire classification on the fire information and generating corresponding alarm signals according to the classified fire classes. According to the multi-factor flame recognition method, the obtained live video frames are processed separately, whether the fire information which can be developed into a fire appears or not in a live monitoring environment is analyzed in real time, fire confirmation is carried out again, whether the fire information really exists or not is further judged, fire class analysis is also carried out on the fire information and alarms are generated, so that fire information recognition is more accurate.

Description

technical field [0001] The present application relates to the technical field of electronic intelligent fire protection, and in particular relates to a multi-factor flame identification method applicable to an embedded platform. Background technique [0002] At present, there are three mainstream methods for fire identification. The first method uses traditional fire detection sensors to detect fire information, which generally has the disadvantage of long detection time and low accuracy. The second is to use image recognition to detect fire information, that is, to use traditional digital image processing methods to manually set the feature dimension of fire for fire identification, that is, to manually design multiple representative features to represent fire information. However, due to the limited characteristics of artificial fire representation, fire information in different scenes or backgrounds cannot be expressed normally, and there are generally disadvantages of hi...

AI Technical Summary

Problems solved by technology

[0002] At present, there are three mainstream methods for fire identification. The first method uses traditional fire detection sensors to detect fire information, which generally has the disadvantage of long detection time and low accuracy.

The second is to use image recognition to detect fire information, that is, to use traditional digital image processing methods to manually set the feature dimension of fire for fire identification, that is, to manually design multiple representative features to represent fire information. However, due to the limited characteristics of artificial fire representation, fire information in different scenes or different backgrounds cannot be expressed normally, and there are generally disadvantages of high misjudgment rate and low robustness.

like figure 1 As shown, it is a diagram of a traditional fire identification system. When a general-purpose deep learning method is used for fire identification, the on-site video collected by multiple video acquisition terminals is transmitted to the background server through the switch, and the background server performs centralized calculation, resulting in calculation Huge amount

Images